Oncogene (1997) 14, 1819 ± 1825  1997 Stockton Press All rights reserved 0950 ± 9232/97 $12.00

Inhibition of activated Ras-induced neuronal di€erentiation of PC12 cells by the LIM domain of LIM-kinase 1

Osamu Higuchi1, Toru Amano1, Neng Yang1 and Kensaku Mizuno1,2

1Department of Biology, Faculty of Science, Kyushu University, Hakozaki, Fukuoka 812-81; 2Inheritance and Variation Group, PRESTO, Japan Science and Technology Corporation, Seika-cho, Kyoto 619-02, Japan

LIM-kinase 1 and 2 (LIMK1 and LIMK2) are members LIMK2 mRNA is expressed in various tissues (Okano of a novel class of protein kinases with structures et al., 1995; Nunoue et al., 1995). Chromosomal composed of two LIM motifs at the N-terminus and an localization of human genes for LIMK1 and LIMK2 unusual protein kinase domain at the C-terminus. The was assigned to 7q11.23 and 22q12, respectively cellular functions of the LIMK family proteins have (Okano et al., 1995). remained unknown. In the present study, we examined LIMK1 and LIMK2 possess unique structural e€ects of LIMKs on neuronal di€erentiation of PC12 features, consisting of two tandemly arrayed LIM pheochromocytoma cells. Transient expression analyses motifs at the N-terminus and an unusual protein kinase revealed that LIMK1, in itself, had no apparent e€ect on domain at the C-terminus. The LIM motif was de®ned as PC12 cells, but the oncogenic Ras-induced di€erentiation a structural motif composed of two adjacent zinc ®ngers of PC12 cells was notably inhibited by co-expression separated by a two-amino-acids linker and with a with LIMK1 or LIMK2. A mutant of LIMK1 lacking a consensus sequence (CX2CX16 ± 23HX2C)X2(CX2CX16 ± 21- protein kinase domain (DK) similarly inhibited Ras- CX2±3C/H/D) (Sa nchez-Garcia and Rabbitts, 1994; induced di€erentiation of PC12 cells, but a mutant Dawid et al., 1995). Nuclear magnetic resonance lacking a LIM domain (DLIM) failed to do so, analysis revealed the three-dimensional structure of the indicating that a LIM domain but not a protein kinase LIM motif; it is constructed of antiparallel b sheets with domain is required for the inhibitory activity. This notion two zinc ion-binding modules (Pe rez-Alvarado et al., was further supported by the ®nding that mutation, 1994). The LIM motif was found in diverse proteins, changing conserved cysteines involved in zinc coordina- including homeodomain-containing transcription fac- tion to glycines in both of two LIM motifs, abolished the tors, cytoskeletal proteins, and other signaling proteins, inhibitory activity of DK. Additionally, we also found which were found to be involved in cell fate determina- that the constitutively activated MAP kinase kinase tion, growth regulation, and oncogenesis (Sa nchez- (MAPKK)-induced di€erentiation of PC12 cells was Garcia and Rabbitts, 1994; Dawid et al., 1995). inhibited by co-expression with DK. Furthermore, DK did Although most zinc ®nger motifs function by binding not inhibit the kinase activity of MAP kinase (MAPK) to speci®c DNA or RNA sequences, several lines of stimulated by MAPKK, when co-expressed in COS7 evidence obtained from studies on various LIM-contain- cells. These ®ndings suggest that LIMK1 inhibits ing proteins suggest that the LIM motifs are involved in neuronal di€erentiation of PC12 cells, through its LIM protein-protein interactions (Schmeichel and Beckerle, domain and by interfering with events downstream of 1994; Feuerstein et al., 1994; Wadman et al., 1994; Wu MAPK activation. and Gill, 1994). Thus, the LIM motifs in LIMKs may function as binding modules to interact with other Keywords: LIM protein; LIM motif; protein kinase; proteins. Recently we found that LIMK1 associated with MAP kinase; neurite outgrowth several isoforms of protein kinase C through the LIM domain, but physiological signi®cance of this interaction remained unknown (Kuroda et al., 1996). In addition to the existence of LIM motifs in the Introduction molecule, the protein kinase domains of LIMK family members are unique in that they contain the We recently identi®ed human, rat and chicken cDNAs characteristic sequence motif DLNSHN in subdomain encoding a novel class of protein kinases, termed LIM- VIB and a highly basic insert between subdomains VII kinases (LIMKs), composed of LIMK1 and LIMK2 and VIII. Phylogenetic analysis of the protein kinase (Mizuno et al., 1994; Ohashi et al., 1994; Okano et al., domains revealed that LIMK1 and LIMK2, along with 1995; Nunoue et al., 1995). The cDNA corresponding another distantly related protein kinase TESK1 to mouse LIMK1 was identi®ed by other groups (Toshima et al., 1995), form a cluster within a serine/ (Bernard et al., 1994; Cheng and Robertson, 1995; threonine kinase family, and they are obviously ProÈ schel et al., 1995). Sequences and domain structures separated from other known protein kinases. of LIMK1 and LIMK2 are closely related with about a Although serine/threonine kinase activity of LIMK1 50% overall amino acid identity, but patterns of and LIMK2 was shown in in vitro kinase assays expression of mRNAs for these kinases di€er; (Okano et al., 1995; Nunoue et al., 1995; Cheng and LIMK1 mRNA is highly expressed in the brain while Robertson, 1995; ProÈ schel et al., 1995), the cellular target proteins for these kinases have heretofore not Correspondence: K Mizuno been determined. Received 12 September 1996; revised 7 January 1997; accepted Since LIMK family kinases have unique structural 7 January 1997 features, they are likely to play speci®c roles in LIMK1 inhibits PC12 cell differentiation OHiguchiet al 1820 previously uncharacterized signaling pathways. We observed in around 80% of b-galactosidase positive observed that induced overexpression of LIMK1 led blue cells (Figure 1b and c and Table 1). On the other to growth retardation of ®broblasts (Higuchi et al., hand, no apparent morphological change was induced 1996), but other cellular functions of the LIMK family in PC12 cells after transfection with the vector plasmid proteins are not well understood. Predominant (pRc-CMV) or the expression plasmid containing expression of LIMK1 in the brain suggests a speci®c haemagglutinin (HA) epitope-tagged LIMK1 (pRc- role of LIMK1 in neurons. We have now examined CMV-LIMK1-HA) plus with the reporter pRc-CMV- e€ects of overexpression of LIMKs on neuronal b-Gal (Figure 1a and d and Table 1). Expression of di€erentiation of rat pheochromocytoma PC12 cells. LIMK1-HA in PC12 cells was con®rmed by indirect As these cells di€erentiate into neuron-like cells in immuno¯uorescence staining with anti-HA antibody response to nerve growth factor or ectopic expression (Figure 1d). Thus, overexpression of LIMK1 by itself of an activated form of Ras or other signaling had no apparent e€ect on PC12 cells. molecules, including Raf, MAP kinase kinase (MAPKK) and MAP kinase (MAPK), they are a LIMK1 and LIMK2 inhibit activated Ras-induced useful system for studying the role of proteins which neuronal di€erentiation of PC12 cells putatively function in cell signaling and di€erentiation (Noda et al., 1985; Bar-Sagi and Feramisco, 1985; We next asked whether LIMK1 or LIMK2 would Wood et al., 1993; Cowley et al., 1994; Fukuda et al., a€ect activated Ras-induced di€erentiation. Cells were 1995). co-transfected with three distinct expression plasmids In our transient expression experiments, we found containing activated Ras (pRc-CMV-RasV12), LIMK1- that ectopic expression of LIMK1 or LIMK2 inhibits neuronal di€erentiation of PC12 cells, induced by oncogenically activated Ras transfection. Interestingly, the activity required to inhibit di€erentiation depen- dend on the LIM domain and not the protein kinase domain of LIMK1. We also provide evidence indicating that LIMK1 inhibits neuronal differentia- tion by interfering with events downstream of MAPK activation.

Results

Overexpression of LIMK1 had no apparent e€ect on PC12 cells Overexpression of the oncogenically activated form of Ras (substituing Gly12 by Val) induces neuronal di€erentiation of PC12 cells. In the present study, we used transient expression assays to examine the e€ects Figure 1 Morphological changes in PC12 cells after transfection V12 of LIMKs on neuronal di€erentiation of PC12 cells. with the pRc-CMV empty vector (a), pRc-CMV-Ras (b and c) or pRc-CMV-LIMK1-HA (d). In (a ± c), PC12 cells were co- When PC12 cells were transfected with the expression transfected with 9 mg of the indicated plasmid plus with 1 mgof plasmid containing constitutively activated Ras (pRc- reporter plasmid, pRc-CMV-b-Gal. After 72 h, cells were stained CMV-RasV12) and the reporter plasmid containing b- with X-Gal to monitor the transfected cells. Shown are galactosidase (pRc-CMV-b-Gal) in a ratio of 9 : 1, and representative photomicrographs (magni®cation, 6400). Arrows indicate b-galactosidase-positive blue cells. In (d), PC12 cells were 72 h later cells were stained with X-Gal, signi®cant transfected with pRc-CMV-LIMK1-HA and 72 h later stained by morphological changes (including extension of neurites, indirect immuno¯uorescence with anti-HA antibody to con®rm enlargement of cell mass and ¯attened shapes) were expression of LIMK1-HA protein

Table 1 Effects of LIMK mutants on Ras-induced differentiation of PC12 cells Morphological change Plasmid DNAa (%, mean+s.d.) Transfected cellsb (A) pRc-CMV 2.1+1.9 110 pRc-CMV-RasV12 79.9+3.1 379 pRc-CMV-LIMK1-HA 1.4+2.4 63 (B) pRc-CMV-RasV12+pRc-CMV-LIMK1-HA 55.4+0.9 360 pRc-CMV-RasV12+pRc-CMV-DLIM-HA 78.4+1.9 767 pRc-CMV-RasV12+pRc-CMV-DK-HA 52.7+1.0 354 pRc-CMV-RasV12+pRc-CMV-1mDK-HA 55.0+1.4 245 pRc-CMV-RasV12+pRc-CMV-2mDK-HA 52.9+1.9 645 pRc-CMV-RasV12+pRc-CMV-dmDK-HA 77.0+1.8 360 pRc-CMV-RasV12+pRc-CMV-LIMK2-HA 53.8+1.2 274 aCells were transfected with a 9:1 ratio of the indicated plasmid : pRc-CMV- b-Gal in (A), and with a 9:3:1 ratio of Ras : LIMK construct : pRc-CMV- b-Gal in (B). bX-Gal positive blue cells were counted. Cell numbers are the sum of three or four independent experiments LIMK1 inhibits PC12 cell differentiation O Higuchi et al 1821 HA (pRc-CMV-LIMK1-HA) and the reporter pRc- a CMV-b-Gal, in a ratio of 9 : 3 : 1 and 72 h later the number of morphologically changed cells in b- galactosidase positive blue cells was scored. The percentage of di€erentiated cells in blue cells was reduced to 55%, compared with the percentage (80%) of di€erentiated cells in case of transfection with activated Ras alone (Table 1). These values were obtained reproducibly in four independent experi- ments. Similarly, co-transfection with the expression plasmid containing LIMK2-HA reduced the number of Ras-induced di€erentiated cells to 54% (Table 1). These observations suggest that both LIMK1 and LIMK2 have an inhibitory e€ect on the activated Ras- induced PC12 cell di€erentiation. Incidentally, in our transient expression assay, it is dicult to examine the e€ects of ectopically expressed proteins on NGF- induced neuronal di€erentiation of PC12 cells.

A LIM but not a kinase domain is required for activity of LIMK1 to inhibit Ras-induced PC12 cell b di€erentiation To determine the functional domain of LIMK1 K-HA essential for activity to inhibit Ras-induced neuronal M.W. ∆ K-HA LIM-HA di€erentiation of PC12 cells, we constructed expression (kDa) Mock LIMK1-HA ∆ dm ∆ LIMK2-HA plasmids containing the LIMK1 mutants lacking a protein kinase domain (DK-HA) or a LIM domain 87.0 — (DLIM-HA) (Figure 2a). Authenticity of the plasmid construction was con®rmed by analysing the sizes of mutant proteins expressed in COS cells; they were in agreement with the predicted molecular size (Figure 44.1 — 2b). Expression of mutant proteins in PC12 cells was con®rmed by indirect immuno¯uorescence staining with anti-HA antibody, in a similar manner shown in Figure 1d. Co-expression of DK-HA, but not DLIM- 32.7 — HA, inhibited Ras-induced neuronal di€erentiation of PC12 cells to an extent similar to that seen with the Figure 2 HA-tagged LIMK1 and LIMK2 constructs. (a) wild-type LIMK1-HA (Table 1). Therefore, the N- Schematic representation of C-terminally HA-tagged LIMK1 terminal LIM region, but not the C-terminal protein and LIMK2 and variously mutated forms of LIMK1. The HA epitope at the C-terminus is indicated by a checked box. The kinase domain, is presumably required for the activity shaded and closed box indicate LIM domain and protein kinase to interfere with Ras-induced di€erentiation of PC12 domain, respectively. Asterisk indicates the site of point mutation, cells. at which the conserved cysteine residue involved in zinc To further de®ne the role of LIM motifs for the coordination was converted to glycine, as shown in lower panel. (b) Immunoblot analysis of HA-tagged LIMK1 and LIMK2 inhibitory activity of LIMK1, a series of point mutant proteins. The cDNA constructs inserted into pUC-SRa mutations changing conserved cysteines to glycines expression vector were transfected into COS7 cells, and the cell were introduced into DK-HA mutant (Figure 2a). lysates were analysed by immunoblotting with anti-HA antiserum. These mutations are expected to destroy the zinc- The elution positions of size marker proteins are indicated on the coordinated tertiary structure of LIM motifs (Taira et left al., 1994). 1mDK-HA and 2mDK-HA are mutants of DK-HA with a point mutation of Cys to Gly at the ®rst and the second LIM motif (LIM1 and LIM2), zinc-coordinated conformation of either LIM1 or respectively. dmDK-HA is a mutant of DK-HA with LIM2 motif is required for the inhibitory activity of double point mutations of Cys to Gly at both LIM1 LIMK1; function of LIM1 and LIM2 motifs is and LIM2 motifs. As summarized in Table 1, when redundant in this activity. transfected into PC12 cells, 1mDK-HA and 2mDK-HA behaved like the wild-type LIMK1-HA and DK-HA, Constitutively activated MAPKK-induced PC12 inhibiting Ras-induced cell di€erentiation. In contrast, di€erentiation is inhibited by DK mutant of LIMK1 dmDK-HA failed to inhibit the Ras-induced differ- entiation of PC12 cells. Thus, mutation leading to When PC12 cells di€erentiate to neuron-like cells, a destruction of either one of the two LIM motifs sequential serine/threonine kinase cascade, including retained the function of DK-HA, but mutation leading Raf, MAPKK, and MAPK, is activated, following to destruction of both LIM motifs resulted in loss of activation of Ras. Cowley et al. (1994) reported that function of DK-HA. These data provide strong expression of the constitutively activated form of evidence for the essential role of LIM motif in the MAPKK induced neuronal di€erentiation of PC12 inhibitory activity of LIMK1 and also indicate that the cells. To determine the point at which LIMK1 inhibits LIMK1 inhibits PC12 cell differentiation OHiguchiet al 1822 the Ras-induced di€erentiation signaling pathway, we MAPK-HA containing HA-tagged MAPK, with or examined the e€ect of overexpression of DK-HA on without the expression plasmids containing activated activated MAPKK-induced di€erentiation of PC12 MAPKK (pUC-SRa-DNSESE) and/or DK mutant of cells. The expression plasmid pRc-CMV-DNSESE LIMK1 lacking the HA epitope (pUC-SRa-DK). After containing Xenopus DNSESE was used for expression 48 h, MAPK-HA proteins were immunoprecipitated of constitutively activated MAPKK, in which the N- with anti-HA antiserum, and the kinase activity of terminal negative regulation domain (residues 32 ± 51) MAPK-HA was measured, using myelin basic protein was deleted and two activating phosphorylation sites (MBP) as a substrate. MAPK-HA showed no MBP (Ser218 and Ser222) were replaced by glutamic acid phosphorylation activity when transfected alone, but residues (Gotoh et al., 1995; Mansour et al., 1994). In was activated by co-transfection with activated the transient transfection assay with the b-Gal reporter MAPKK (DNSESE) (Figure 4a, lanes 1 and 2). plasmid, DNSESE induced neuronal di€erentiation of Expression of an excess amount of DK mutant of PC12 cells in 57% of blue cells (Figure 3). When the LIMK1 in COS cells did not reduce the activity of DK-HA mutant of LIMK1 was co-transfected with MAPK-HA induced by DNSESE (Figure 4a, lane 3), DNSESE, the number of di€erentiated cells in blue hence, the LIM domain of LIMK1 did not seem to cells decreased to near 30%; co-expression of dmDK- interfere with the MAPKK-induced activation of HA had no such e€ect. Thus, DK-HA but not dmDK- MAPK. DK probably inhibits neuronal di€erentiation HA speci®cally inhibited the activated MAPKK- of PC12 cells further downstream of MAPK activation. induced di€erentiation of PC12 cells, similar to their In addition, the observation that expression level of actions on the activated Ras-induced di€erentiation. MAPKK (DNSESE) was not changed by overexpres- These ®ndings suggest that DK-HA inhibits the Ras- sion of DK (Figure 4c, lanes 2 and 3) suggests that the induced signaling pathway downstream of MAPKK inhibitory e€ect of DK on neuronal di€erentiation is activation. not due to the reduction of the level of MAPKK protein. Lack of e€ect of the DK mutant on MAPK activation by MAPKK Discussion As the data described above indicated that DK-HA inhibited neuronal di€erentiation of PC12 cells down- Several lines of evidence indicate that sustained stream of MAPKK activation, we next determined activation of Ras-MAPK pathway correlates well with whether or not DK-HA would inhibit the activation of neuronal di€erentiation of PC12 cells (Marshall, 1995). MAPK, which is expected to be induced by activated MAPKK. COS7 cells were transfected with pUC-SRa- 1 2 3

a MBP

b MAPK-HA

∆NSESE c

d ∆K

MAPK-HA + + + ∆NSESE – + + ∆K – – + Figure 4 E€ects of overexpression of DK on MAPK activation by DNSESE. COS7 cells were co-transfected with pUC-SRa- MAPK-HA, pUC-SRa-DNSESE and pUC-SRa-DK (which have no HA epitope), as indicated. Ratio of plasmids of pUC-SRa- Figure 3 Inhibition of activated Ras or activated MAPKK- MAPK-HA, pUC-SRa-DNSESE and pUC-SRa-DK was induced di€erentiation of PC12 cells by DK mutant, but not 1 : 10 : 100. After 48 h, MAPK-HA was immunoprecipitated with dmDK mutant of LIMK1. PC12 cells were transfected with the anti-HA antiserum. The immunoprecipitates were incubated with expression plasmid (9 mg) containing constitutively activated form MBP and [g-32P]ATP for 15 min at 308C, and separated on 15% of Ras or MAPKK (pRc-CMV-RasV12 or pRc-CMV-DNSESE), SDS-polyacrylamide gel. Proteins were transferred onto PVDF in the presence or absence of the plasmid (3 mg) containing membrane, and phosphorylated MBP was detected by autoradio- LIMK1 mutant (pRc-CMV-DK-HA or pRc-CMV-dmDK-HA). graphy (a). Expression of MAPK-HA, DNSESE and DK was Cells were co-transfected with 1 mg of pRc-CMV-b-Gal and 72 h detected by immunoblotting with anti-HA antiserum (b), anti- later stained with X-Gal to monitor the transfected cells. The MAPKK antiserum (c), or anti-LIMK1 antiserum (d). Lane 1, percentage of morphologically changed cells in b-galactosidase pUC-SRa-MAPK-HA alone; lane 2, pUC-SRa-MAPK-HA and positive blue cells was counted. Data are the mean+s.d. of three pUC-SRa-DNSESE; lane 3, pUC-SRa-MAPK-HA, pUC-SRa- to four separate experiments DNSESE and pUC-SRa-DK LIMK1 inhibits PC12 cell differentiation O Higuchi et al 1823 Expression of the constitutively activated form of Ras, conserved cysteine residues that coordinate zinc ion by Raf, MAPKK or MAPK by microinjection, infection glycine residues in LIM1 and LIM2 motif of DK using a viral vector or transient transfection leads to mutant of LIMK1 suggest that the zinc-coordinated development of neurites in PC12 cells (Noda et al., structure of either one of the two LIM motifs is 1985; Bar-Sagi and Feramisco, 1985; Wood et al., required to exhibit inhibitory activity. Based on the 1993; Cowley et al., 1994; Fukuda et al., 1995). Using signi®cant di€erence in amino acid sequences of LIM1 the transient expression system, we found that the and LIM2 motifs in LIMK1, the respective LIM motifs activated Ras-induced PC12 cell di€erentiation was likely have distinct target binding speci®cities, as inhibited by co-expression of LIMK1 or LIMK2. previously noted for LIM motifs in zyxin and Based on ®ndings that the activated MAPKK-induced enigma, which contain multiple LIM motifs with di€erentiation of PC12 cells was also inhibited by the distinct binding speci®cities in the single molecule DK mutant of LIMK1 and that DK had no e€ect on (Schmeichel and Beckerle, 1994; Wu and Gill, 1994; MAPKK-stimulated MAPK activation in COS cells, Durick et al., 1996). Thus, the respective LIM motifs in we suggest that LIMK1 inhibits PC12 cell differentia- LIMK1 may function by binding with distinct sites of tion, probably by interfering with events downstream the single target protein or distinct proteins that are of MAPK activation. involved in Ras-induced di€erentiation signaling. To Despite recent advances in delineating signaling better understand mechanisms by which LIMK1 pathways, little is known of downstream events of inhibits Ras-induced di€erentiation signaling and MAPK activation. MAPK, upon activation, translo- mechanisms of the functional redundancy of two cates to the nucleus, phosphorylates and activates LIM motifs in LIMK1, further identi®cation and transcription factors, some of which will lead to the characterization of proteins interacting with the expression of genes required for neuronal differentia- individual LIM motifs of LIMK1 will need to be done. tion of PC12 cells (Marshall, 1995; Nishida and Gotoh, Recently Keating's group reported that hemizygotic 1993; Blenis, 1993; Davis, 1993). In addition to shifts in deletion of the LIMK1 gene leads to impaired gene expression, MAPK activation also induces visuospatial constructive cognition in patients of dynamic cytoskeletal rearrangements leading to mor- Williams' syndrome (Frangiskakis et al., 1996). This phological changes, such as an increase of cell mass cognitive function is likely supported by a network of and neurite outgrowth, in PC12 cells. Since LIMKs are neurons capable of parallel processing. Together with mainly localized in the cytoplasm (see Figure 1d and the ®nding that LIMK1 is predominantly expressed in Okano et al., 1995), they a€ect PC12 cell di€erentiation the brain of rat embryo (Mori et al., 1996), LIMK1 probably by interfering with the events related to seems to play an important role in the development of cytoskeletal reorganization, although the possibility the central nervous system. The ®nding of the that LIMKs perturb the nuclear events controlling inhibitory e€ect of LIM domain of LIMK1 on gene expression cannot be excluded at present. neuronal di€erentiation of PC12 cells in this study In this respect, it is noted that several LIM- will serve as the ®rst step to explore the physiological containing proteins, including paxillin, zyxin and functions of LIMKs in neuronal development. cysteine-rich protein (CRP), are associated with cytoskeletal structures. Paxillin binds to cytoskeletal component proteins, such as vinculin, talin and Materials and methods pp125FAK, and zyxin binds to a-actinin, through the regions outside of the LIM domains (Crawford et al., Plasmid construction 1992; Turner and Miller, 1994; Salgia et al., 1995), The full-length coding sequence of human LIMK1 cDNA while zyxin and CRP are associated with each other (3029 base pairs), obtained by ApaI±HindIII digestion through LIM ± LIM interactions (Sadler et al., 1992; of the clone S9-14 (Mizuno et al., 1994), was inserted Schmeichel and Beckerle, 1994). As the LIM domains into pBluescript SKII7 (Stratagene) with a NotI linker are thought to function as the interface in protein- (pBS-LIMK1). To generate the plasmid encoding LIMK1- protein interactions, the LIM domains of cytoskeletal HA, C-terminally tagged with epitope peptide (YPYDVP- LIM proteins are likely involved in their binding to DYA) of in¯uenza virus haemagglutinin (HA), the cDNA cytoskeletal component proteins or regulatory proteins, fragment was ampli®ed by polymerase chain reaction and thus may contribute to the regulation of (PCR), using the upper primer 5'-GACAGCCAGTA CCCA-3' and the lower primer 5'-GCTCTAGAGG- cytoskeletal reorganization. Similarly, the LIM do- CCTCAGGCCATAGTCGGGGACGTCATAGGGGGGT- main of LIMKs may be involved in the regulation of AAGGCAGTCCGCTCTCGCC-3' (containing an HA the cytoskeletal reorganization by binding to cytoske- epitope, a stop codon and XbaI site). The PCR-ampli®ed letal LIM proteins or other cytoskeletal component or fragment was digested with SphIandXbaI and ligated into regulatory proteins and consequently a€ect on the the SphI, XbaI-digested pBS-LIMK1, to replace the morphological changes of PC12 cells. Very recently we original SphI±XbaI fragment of LIMK1 cDNA with the found that LIMK1 associates with various isoforms of HA-tagged fragment. The resulting plasmid was cut with protein kinase C through the LIM domain (Kuroda et NotI and the insert was subcloned into the expression al., 1996). As protein kinase C isoforms are known to vector pRc-CMV (Invitrogen) or pUC-SRa vector (Takebe bind to and phosphorylate several cytoskeletal proteins et al., 1988) to produce pRc-CMV-LIMK1-HA or pUC- SRa-LIMK1-HA. To construct the expression plasmid (Nishizuka, 1986; Blobe et al., 1996), the LIM domain pRc-CMV-DLIM-HA coding for DLIM-HA lacking a of LIMK1 may regulate the cytoskeletal reorganization LIM domain of LIMK1-HA, pRc-CMV-LIMK1-HA was through its interaction with protein kinase C. digested with BalIandEcoO65I and ligated with a HindIII Mutation analysis revealed that the LIM domain but linker. To construct the plasmid pRc-CMV-DK-HA coding not the protein kinase domain is required for the for DK-HA lacking a kinase domain of LIMK1-HA, pRc- inhibitory function of LIMK1. Point mutations of CMV-LIMK1-HA was digested with PstIandSphIandthe LIMK1 inhibits PC12 cell differentiation OHiguchiet al 1824 PstI±SphI fragment within the LIMK1 sequence was cells were transfected with the expression plasmids, as removed. Point mutations in the LIM domain of LIMK1 described above. After 72 h, cells were ®xed with 4% were introduced, using mutated oligodeoxynucleotides and paraformaldehyde for 15 min, permeabilized with 0.2% an in vitro site-directed mutagenesis kit (Clontech). The b- Triton X-100 in PBS for 15 min, and quenched in 100 mM galactosidase cDNA was excised out from pSV-b-galacto- glycine in PBS for 15 min, then in 10% fetal calf serum in sidase (Promega) by HindIII ± ApaI digestion, and sub- PBS for 10 min. The cells were then incubated for 2 h at cloned into the pRc-CMV vector (pRc-CMV-b-Gal). The room temperature with anti-HA epitope rabbit antisera c-Ha-RasV12 cDNA was cut out from pMT-cHr-2 (provided (HA.11, Babco) at a dilution of 1 : 1000 in PBS, and then by Dr S Hattori) (Hattori et al., 1985) by BglII digestion, for 1 h with ¯uorescein isothiocyanate (FITC)-conjugated and subcloned into pRc-CMV with a NotI linker (pRc- goat anti-rabbit IgG antibody (Zymed) at a dilution of CMV-RasV12). SRa456 -XenopusD(32 ± 51) - S218E/S222E- 1 : 10 000 in PBS. Fluorescent cells were photographed MAPKK (DNSESE, provided by Drs Y Gotoh and E using a epi¯uorescent microscope (Zeiss). Nishida) (Gotoh et al., 1995) was digested with BglII and the BglII sites in both ends were substituted by NotIsites. Immunoprecipitation and immunoblotting The NotI-digested cDNA was subcloned into pRc-CMV or pUC-SRa vector to generate pRc-CMV-DNSESE or pUC- Immunoprecipitation and immunoblotting were carried out SRa-DNSESE. SRa456-Xenopus MAPK expression vector as described previously (Okano et al., 1995), except that (provided by Drs Y Gotoh and E Nishida) (Fukuda et al., anti-HA epitope polyclonal antisera (HA.11) was used for 1995) was digested by BamHI and the BamHI sites immunoprecipitation and for the primary antibody for in both ends were substituted by NotI sites. The MAPK immunoblotting. Anti-MAPKK antibody was provided by cDNA was HA epitope-tagged at the C-terminus by Drs Y Gotoh and E Nishida. PCR ampli®cation, using the 5' primer (5'-TCCCAATG- CAGATCCC-3')andthe3'primer (5'-GTGCGGCC Assay for MAPK activity GCTGTTCAGGCATAGTCGGGGACGTCATAGGGGT- ACCCTGGCTGGAATCT-3') and subcloning into pUC- Subcon¯uent COS7 cells were transfected with expression SRa vector (pUC-SRa-MAPK-HA), in a similar manner. plasmids containing DKordmDK(pUC-SRa-DKor The authenticity of these expression plasmids was -dmDK), activated MAPKK (SRa-NSESE), and HA- con®rmed by nucleotide sequence analysis. tagged MAPK (pUC-SRa-MAPK-HA) in a ratio of 100 : 10 : 1, using the liposome methods described above. Total amount of plasmid DNA was adjusted to 22.2 mgper Transient expression assay for neuronal di€erentiation of PC12 100 mm culture plate with a pUC-SRa vector plasmid. cells Two days later, the cells were washed with PBS and lysed Transient expression omit analysis of neuronal differentia- on ice with NP-40 lysis bu€er (50 mM HEPES, pH 7.4,

tion of PC12 cells was carried out as described (Teng et al., 150 mM NaCl, 1% Nonidet P-40, 1 mM Na3VO4,50mM 1995). PC12 cells obtained from Japanese Cancer Research NaF) containing 0.05% SDS. MAPK-HA was immuno- Resources Bank (JCRB) were cultured in collagen-coated precipitated by incubation of the lysates with anti-HA 100 mm dish in Dulbecco's modi®ed Eagle's medium antisera for 1 h and then with Protein A-Sepharose for 1 h. (DMEM) supplemented with 10% horse serum (GIBCO- The immune complex was washed twice with washing BRL) and 5% fetal calf serum (FCS). After 24 h, cells were bu€er (20 mM HEPES, pH 7.4, 150 mM NaCl, 0.5% V12 transfected with 9 mgofpRc-CMV-Ras ,3mgofpRc- Nonidet P-40, 1 mM Na3VO4,50mMNaF) and twice with CMV-LIMK derivatives, and 1 mgofpRc-CMV-b-Gal, kinase reaction bu€er (50 mM HEPES, pH 7.4, 1 mM

which were mixed with 65 mg of liposome (provided by Dr dithiothreitol, 5 mM MgCl2,1mM Na3VO4,1mM NaF). AIto)(Itoet al., 1990) in Hepes-based saline bu€er MAPK activity was determined by incubation for 15 min (pH 7.4). 72 h after transfection, cells were washed with at 308C with 2.5 mM [g-32P]ATP (5000 ± 10 000 c.p.m./pmol, phosphate-bu€ered saline (PBS) and ®xed with 1% Amersham) and 5 mg of myelin basic protein (MBP) in glutaraldehyde for 5 min at 48C. After washing twice with 20 ml of kinase reaction bu€er. Samples were mixed with

PBS containing 5 mM MgCl2, cells were stained by 20 mlof26Laemmli's bu€er containing 100 mM DTT, incubation for over 3 h at 378CinsolutionofPBS boiled for 10 min and analysed by SDS electrophoresis on

containing 20 mM K3Fe(CN)6,20mM K4Fe(CN)6,1mM 15% polyacrylamide gels. Proteins were transferred to 32 MgCl2, and 1 mg/ml X-Gal (5-bromo-4-chloro-3-indolyl-b- PVDF membranes, and P-labeled MBP was detected by D-galactoside). The b-galactosidase positive blue cells were autoradiography with the aid of intensifying screens. scored by phase-contrast microscopy. Morphologically altered cells were assessed by neurite outgrowth with a ¯attened shape and increase in body mass, as typi®ed in Acknowledgements Figure 1b and c. We thank Dr Y Fujiki for encouragement and M Ohara for comments. We also thank Drs S Hattori, E Nishida and Y Gotoh for kind gifts of plasmids and antibodies. This Immuno¯uorescence work was supported by grants from the Japan Science and PC12 cells were plated on 10 mm glass coverslips coated Technology Corporation and the Ministry of Education, with poly-D-lysine, collagen, and polyethylenimine. These Science, Sports and Culture of Japan.

References

Bar-Sagi D and Feramisco JR. (1985). Cell, 42, 841 ± 848. Cowley S, Paterson H, Kemp P and Marshall CJ. (1994). Bernard O, Ganiatsas S, Kannourakis G and Dringen R. Cell, 77, 841 ± 852. (1994). Cell Growth Di€er., 5, 1159 ± 1171. Crawford AW, Michelsen JW and Beckerle MC. (1992). J. Blenis J. (1993). Proc. Natl. Acad. Sci. USA, 90, 5889 ± 5892. Cell Biol., 116, 1381 ± 1393. Blobe GC, Stribling DS, Fabbro D, Stabel S and Hannum Davis R. (1993). J. Biol. Chem., 268, 14553 ± 14556. YA. (1996). J. Biol. Chem., 271, 15823 ± 15830. Dawid IB, Toyama R and Taira M. (1995). C. R. Acad. Sci. Cheng AK and Robertson EJ. (1995). Mech. Dev., 52, 187 ± Paris, 318, 295 ± 306. 197. LIMK1 inhibits PC12 cell differentiation O Higuchi et al 1825 Durick K, Wu R-Y, Gill GN and Taylor SS. (1996). J. Biol. OkanoI,HiraokaJ,OteraH,NunoueK,OhashiK,Iwashita Chem., 271, 12691 ± 12694. S, Hirai M and Mizuno K. (1995). J. Biol. Chem., 270, Feuerstein R, Wang X, Song D, Cooke NE and Liebhaber S. 31321 ± 31330. (1994). Proc. Natl. Acad. Sci. USA, 91, 10655 ± 10659. ProÈ schel C, Blouin M-J, Gutowski NJ, Ludwig R and Noble Frangiskakis JM, Ewart AK, Morris CA, Mervis CB, M. (1995). Oncogene, 11, 1271 ± 1281. Bertrand J, Robinson BF, Klein BP, Ensing GJ, Everett Pe rez-Alvarado GC, Miles C, Michelsen JW, Louis HA, LA, Green ED, Prochel C, Gutowski NJ, Noble M, Winge DR, Beckerle MC and Summers MF. (1994). Atkinson DL, Odelberg SJ and Keating MT. (1996). Cell, Nature Struct. Biol., 1, 388 ± 398. 86, 59 ± 69. Sadler I, Crawford AW, Michelsen JW and Beckerle MC. Fukuda M, Gotoh Y, Tachibana T, Dell K, Hattori S, (1992). J. Cell Biol., 119, 1573 ± 1587. Yoneda Y and Nishida E. (1995). Oncogene, 11, 239 ± 244. Salgia R, Li J-L, Lo SH, Brunkhorst B, Kansas GS, Sobhany Gotoh Y, Masuyama N, Dell K, Shirakabe K and Nishida E. ES, Sun Y, Pisick E, Hallek M, Ernst T, Tantravahi R, (1995). J. Biol. Chem., 270, 25898 ± 15904. Chen LB and Grin JD. (1995). J. Biol. Chem., 270, Hattori S, Halliday K, Ulsh LS and Shih TY. (1985). Mol. 5039 ± 5047. Cell. Biol., 5, 1449 ± 1455. Sa nchez-GarcõÂa I and Rabbitts TH. (1994). Trends Genet., Higuchi O, Baeg G-H, Akiyama T and Mizuno K. (1996). 10, 315 ± 320. FEBS Lett., 396, 81 ± 86. Schmeichel KL and Beckerle MC. (1994). Cell, 79, 211 ± 219. Ito A, Miyazoe R, Mitoma J, Akao T, Osaki T and Kunitake Taira M, Otani H, Saint-Jeannet J-P and Dawid IB. (1994). T. (1990). Biochem. Int., 22, 235 ± 241. Nature, 372, 677 ± 679. Kuroda S, Tokunaga C, Kiyohara Y, Higuchi O, Mizuno K, Takebe Y, Seiki M, Fujisawa J, Hoy P, Yokota K, Arai K, Gill GN and Kikkawa U. (1996). J. Biol. Chem., 271, Yoshida M and Arai N. (1988). Mol. Cell. Biol., 8, 466 ± 31029 ± 31032. 472. Mansour SJ, Matten WT, Hermann AS, Gandiaa JM, Rong Teng KK, Lander H, Fajardo JE, Hanafusa H, Hempstead S, Fukasawa K, Vande Woude GF and Ahn NG. (1994). BL and Birge RB. (1995). J. Biol. Chem., 270, 20677 ± Science, 265, 966 ± 970. 20685. Marshall CJ. (1995). Cell, 80, 179 ± 185. Toshima J, Ohashi K, Okano I, Nunoue K, Kishioka M, Mizuno K, Okano I, Ohashi K, Nunoue K, Kuma K, Miyata Kuma K, Miyata T, Hirai M, Baba T and Mizuno K. T and Nakamura T. (1994). Oncogene, 9, 1605 ± 1612. (1995). J. Biol. Chem., 270, 31331 ± 31337. Mori T, Okano I, Mizuno K, Tohyama M and Wanaka A. Turner CE and Miller JT. (1994). J. Cell Sci., 107, 1583 ± (1996). Molec. Brain Res., in press. 1591. Nishida E and Gotoh Y. (1993). Trends Biochem. Sci., 18, Wadman I, Li J, Bash RO, Forster A, Osada H, Rabbitts TH 128 ± 131. and Baer R. (1994). EMBO J., 13, 4831 ± 4839. Nishizuka Y. (1986). Science, 233, 305 ± 312. Wood KW, Qi H, D'Arcangelo G, Armstrong RC, Roberts Noda M, Ko M, Ogura A, Liu D, Armano T, Takano T and TM and Halegous S. (1993). Proc. Natl. Acad. Sci. USA, Ikawa Y. (1985). Nature, 318, 73 ± 75. 90, 5016 ± 5020. Nunoue K, Ohashi K, Okano I and Mizuno K. (1995). Wu R-Y and Gill GN. (1994). J. Biol. Chem., 269, 25085 ± Oncogene, 11, 701 ± 710. 25090. Ohashi K, Toshima J, Tajinda K, Nakamura T and Mizuno K. (1994). J. Biochem., 111, 636 ± 642.